Transcript Empirical Analytical Science

北京师范大学研究生课程
教育研究的基础：方法论、知识论及本体论
Topic 2
Methodological & Epistemological Foundations of
Empirical Analytical Sciences
http://www.fed.cuhk.edu.hk/~wktsang/
The Methodological & Epistemological
Assumptions of Positivism
According to the traditions passed on by the
Vienna Circle to the approach of empiricalanalytical sciences in the late twentieth
century, positivism, more specifically empiricalor logical-positivism, general works with the
following methodological and epistemological
assertions (Byant, 1985; Giddens, 1974;
Kolakowski, 1966):
The Methodological & Epistemological
Assumptions of Positivism
…..
Empiricism: The positivism approach believes that
“there is but one world, and it has an objective
existence.”(Byrant, 1985, P. 12) And this external
world can only be experience by human sensoryobservation. Accordingly, “we are entitled to record
only that which is actually manifested in
experience.” (Kolakowski, 1966, P. 11).
The Methodological & Epistemological
Assumptions of Positivism
…..
Atomism or norminalism: As Kolakowski states,
“We may not assume that any insight formulated to
general terms can have any real referent other than
individual facts.” (1966, P. 15) Accordingly, for every
abstract science, their task is to distill and abridge
the recorded experiences to concrete and actual
individual units, which may be in the form of
element, atom, DNA, or in social science individual.
As a result, such an assertion has been termed in
social science as methodological individualism.
The Methodological & Epistemological
Assumptions of Positivism
…..:
Value free: Based on the atomistic and empiricistic
assumption, positivists assert they should state their
findings and results with a plain and simple
language and most importantly they should restraint
form adding any value implications and/or
judgements to their findings and results.
Conviction of nomology: Following their empiricist,
abstract-atomist, and free-value assertions,
positivists therefore assume that they can and
should abstract some “law-like generalization” from
the external world. In other words, they are after
permanent and universal nomology.
The Methodological & Epistemological
Assumptions of Positivism
…..
Unity of scientific method: Given the above
convictions, positivist therefore demand all scientific
research should comply with the same method of
enquiry practice and demonstrated by positivists of
logical-empiricism.
Carl G. Hempel’s General-Law
Explanatory Model
In 1942, Carl Hempel, Professor of Princeton
University, published an article in Journal of
Philosophy entitle “The Function of General
Law in History”. The primary objective of the
article is to reinstate the Vienna-Circle’s
advocation of “the methodological unity of
empirical science.” (Hempel, 1965/1942, P. 243)
Carl G. Hempel’s General-Law
Explanatory Model
Taking history (as well as other social
sciences) as an empirical science, Hempel
began his article with an outright criticism on
the historians by underlining that they have
focused their research efforts on “the
description of particular events of the past”
rather than on “the search for general laws
which might govern those events” and “it is
certainly unacceptable”. (Hempel, 1965, P. 231)
Hempel then explicate in details his explanatory
model by general law.
Carl G. Hempel’s General-Law
Explanatory Model
Deductive-nomological model: The ideal-typical
model of explanation in empirical sciences
(including both natural and social sciences) is
what he characterizes as deductivenomological model. The model is made up of
three parts.
Carl G. Hempel’s General-Law
Explanatory Model
Deductive-nomological model: …
The model is made up of three parts.
The specific events (E) to be explained, in Hempel’s
terms, the explanandum
The possible antecedent condition (C) which brings
about the occurrence of that specific events
In order to logically deduce a causal relation
between E and C, a general law covering the
occurrences of both E and C is needed. And by
general law it means a statement of universal causalconditions that has been empirically or logical
confirmed. More specifically, causal conditions may
be further differentiated into three types:
Deductive-Nomological Explanatory Model
C1,C2,…Ck
Logical
deduction
L1,L2,,...Lr
E
Statements if
antecedent
conditions
Explanans
General Laws
Description of the
empirical
phenomenon to
be explained
Explanandum
Carl G. Hempel’s General-Law
Explanatory Model
Deductive-nomological model: …
The model is made up of three parts….
….More specifically, causal conditions may be
further differentiated into three types:…
 Sufficient conditions: It refers to the kinds of conditionality
between the C and E, in which the C can exhaustively but
not universally explain the truth of the E.
 Necessary conditions: It refers to the kinds of conditionality
between the E and C, in which the C can universally but not
exhaustively explain the truth of the E.
 Sufficient and necessary conditions: It refers to the kinds of
conditionality between the E and C, in which the C can both
exhaustively and universally explain the truth of the E.
Carl G. Hempel’s General-Law
Explanatory Model
The compromised model: StatisticalProbabilistic explanation:
The statistical-probabilistic model is the type of
explanation commonly use in quantitative
researches in social sciences. It is also made up of
three parts similar to those in nomological-deductive
explanation. There are two differences in
probabilistic explanation. …..
Carl G. Hempel’s General-Law
Explanatory Model
The compromised model: StatisticalProbabilistic explanation:
The statistical-probabilistic model ….
One is that the explanatory premises is not in the
form of law-like / nomological statement of the
sufficient and necessary conditions of the truth of
the explanandum but only a probabilistic statement
specifying the likelihood of the causal relationship
between the explanans and explanandum. The
second difference is that in the conclusion, the
specific explanandum under study cannot be
exhaustive explained by the explanans but can only
be explained in probabilistic terms.
Statistical-Probabilistic Explanatory Model
C1,C2,…Ck
Logical
deduction
L1,L2,,...Lr
E
Statements if
antecedent
conditions
Explanans
Statistical Laws
Description of the
empirical
phenomenon to
be explained
Explanandum
The Constitution of the AnalyticalEmpirical Approach
Transplanting the Deductive-nomological
explanatory model from the natural science,
researchers in social sciences began to
constitute their “Language of social research”
(Lazarsfeld, et al., 1955, see also 1972). As a
result, the methodology of quantitative
research (at least in America) has been
founded. Apart from the deductive-nomological
explanatory model, two of the other principles
of this methodological foundation are the
analytical approach and empiricism.
The Constitution of the AnalyticalEmpirical Approach
….two other methodological foundation are the
analytical approach and empiricism. …
 Analytical approach: By analytical approach, it refers
the way social researchers approach their objects
under study, i.e. the social world or social reality, by
decomposing the objects into elements or properties
and then try to establish causal relations among
them. …
The Constitution of the AnalyticalEmpirical Approach
 Analytical approach: …
In Paul Lazarsfeld own words, “No science deals
with its objects of study in their concreteness. It
selects certain of their properties and attempts to
establish relations among them. The finding of such
laws is the ultimate goal of all scientific inquiries. But
in social sciences the singling out of relevant
properties is in itself a major problem. No standard
terminology has yet been developed for this task.
The properties are sometimes called aspects or
attributes, and often the “variable” is borrowed from
mathematics as most general category.” (Lazarsfeld,
1955, P. 15)
The Constitution of the AnalyticalEmpirical Approach
 Empiricism: It refers to the way social research
approach their evidences to be accepted in verifying
the propositions or more specifically hypotheses.
That is, only observations verified by sensory
experience can be accepted as evidences.
Furthermore, quantitative social researchers would
demand those sensory-observations should be
repeatable, recordable, and quantifiable into a
mathematical “variables”.
Methodological Designs in Social
Inquiries
Experimental design: Based on these three
principles, viz. deductive- nomological
explanatory model, analytical approach, and
empiricism; quantitative social researchers
could only design their studies, which is to
verify causals relation among variables, by
observing and recording the outcomes of their
experimental design. The most typical design
of a true experiment can be represented as
follows (Campbell and Stanley, 1963, Pp. 13-22)
Methodological Designs in Social
Inquiries
Experimental design:
R
R
O1
O3
X
O2
O4
Connotations:
R=Random assignment of subjects
O=Observation
X=Treatment (effects under study)
Methodological Designs in Social
Inquiries
Social-survey design: However, by the very
nature of the social world, it is basically
infeasible or even unethical to conduct
experiments on human subjects. As a result,
most of the social researchers can only
conduct their observation in naturalistic
environment of the social world. That is by
means of social survey. As a result, one of the
three primary design tools, i.e. treatment
manipulation, has to be abandoned. What is left
are the design-tools of one-short observation
and random sampling.
The Underlying Assumptions of
Research Designs in Social Inquiries
Given the principles and operations of the
methodology of quantitative researches, they
reveal that the causal relations between
variables, which quantitative researches are
supposed to have verified, are, in fact, implicated
under at least the following three assumptions
(Hirschi and Selvin, 1996)
The Underlying Assumptions of
Research Designs in Social Inquiries
…implicated under at least the following three
assumptions (Hirschi and Selvin, 1996)
 Assuming covariance as causation: In most of the
social experiments and most notably social surveys,
the observations between the cause-variable and the
effect-variable could only be in association formats.
That is the two sets of variables vary concurrently. It
could not reveal the real causal interactions between
the two set of social properties simply by
observations. As a result two other working
assumption are needed.
The Underlying Assumptions of
Research Designs in Social Inquiries
…implicated under …three assumptions….
Assuming or manipulating the temporal orderings
between the variables: One of the logical conditions
of a causation is that the cause must precede the
effect. And simple statistical covariance can never
substantiate the temporal orderings between two
sets of variables. As a result, temporal orderings
could only be built in by experimental design, that is,
to have the designed treatment carried out right
between the pretest and post-test. However, as for
one-short social survey, the temporal orderings
among various variables could only be assumed. …
The Underlying Assumptions of
Research Designs in Social Inquiries
…implicated under …three assumptions …
 Assuming or manipulating the temporal orderings
between the variables: ….
For examples, in most educational surveys, it is
assumed that parents’ education, socio-economic
status, students’ genders are precedent to students’
educational achievement.
The Underlying Assumptions of
Research Designs in Social Inquiries
…implicated under …three assumptions …
 The assumptions of ceteris paribus: It refers to the
assumption that all other possible factors in the
causal relation are being equal or unchanged. The
only design-tools available for social researchers
deal with this assumption is randomization. In
experimental design, it is the random assignments of
subjects into treatment and non-treatment group. As
for social survey, researchers could only fall back on
to random sampling.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:
Transplanting the logical/empirical positivism
prevailing in natural scientific enquiry, the
analytical-empirical science presumes that
there is no essential difference between natural
and social worlds. As a result, it presumes the
following features upon the social reality it is to
enquire.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:
Objective-analytical approach: By objectiveanalytical approach, it refers to the working
assumption that social scientists impose upon their
objects under study. By objective approach, they
assume that the social world and social reality are
objective facts and most specially “things” external
to and independent of human minds and ideas. ….
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:
Objective-analytical approach: ….
By analytical approach, they further assume that
the social reality can be decomposed into elements
or properties. They further assume that these
elements are causally related into structure of
antecedent causes and consequent effects.
Accordingly, the task of social researchers is to
select the most significant and relevant
components in social reality and formulate the
respective causal proposition (hypothesis) between
them, and then test it against the objectively
existing factual properties found in the world.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:…
Empirical-logical grounded: It refers to the
existential assumption that social researcher
presume upon the social world, from which social
researchers are supposed to find evidences in
verifying or falsifying their propositions or more
specifically hypotheses. By empirical, it refers to the
assumption that features and properties of social
reality are observable by sensory experiences.
Moreover, these evidences collected by sensory
observations can then be test against the respective
hypothesis for verification or falsification. …
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:…
Empirical-logical grounded: …
By logical, it refers to the fact that scientists (mostly
natural scientists, like mathematical physicists) can
also be verified their propositions by formal logical
and mathematical operations. Taken together these
principles of the empirical- and logical-positivism,
they have formed the foundations of the quantitative
research method of social and educational research.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:…
The objective truth and the correspondence
principle of truth: Given the assumptions and
operations of the empirical- and logical-positivism,
the theory of truth that social and educational
researchers should observe is the correspondence
principle of truth. It implies that for a proposition or
belief to be true, its content should find its
corresponding fact in the respective reality.
Accordingly, such conception of truth can be
perceived as objective truth because it can be
verified with objective facts and therefore its validity
is independent of the subjectivity of human mind.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumptions of the object of inquiry:…
Universal and permanent nomology in features: In
correspond with natural science, social scientists
assume that phenomena in social reality persistent
and permanent in features. They are also universally
the same across time and space as natural
phenomena.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumption of the knowledge constituted:
The knowledge to be constituted and verified
by social scientists in analytical-empirical
science are presumed to bear the following
features:
They are coherent sets of verbal propositions (or
even numerical formula) describing the nomological
features of a specific domain the social world.
These verbal propositions are objective statements
in form. That is, they are interest-neutral and valuefree verbal statements recording the objective
features of the social phenomena in point.
The Epistemological Foundation of
Empirical-Analytical Sciences
The assumption of the knowledge constituted:
....
These objective statements of social phenomena, as
a rule, must be verified with relevant empirical
evidences. They must comply with the so-called
“correspondence principle”, that is, the descriptions
in the statements must find its correspondent
evidences in the empirical world.
The Epistemological Foundation of
Empirical-Analytical Sciences
 The implied knowledge-constitutive human
interest: Given its causal-law structure and
universal and permanent features, analyticalempirical knowledge can be used for prediction of
future events. It can be technically applied for
enhancing the occurrence of desirable outcomes
and for preventing undesirable events from
happening. According to Habermas’ formulation,
the knowledge-constitutive human interest that
analytical-empirical knowledge is to serve is the
“technical-cognitive interest” embedded in “work”
in general and material production in particular.
(Habermas, 1971, P. 196).
The Epistemological Foundation of
Empirical-Analytical Sciences
 The implied knowledge-constitutive human
interest: …..
With well work-out analytical-empirical knowledge
or what Habermas termed “technically exploitable
knowledge” (Habermas, 1971, P. 191), human
beings are supposed to be able to control how the
social world works. It can help human to
“engineer” and “domineer” the social world.
Topic 2
Methodological & Epistemological Foundations of
Empirical Analytical Sciences
End
40